For most of a decade, Merck kept running the same experiment on prostate cancer, and prostate cancer kept winning. Checkpoint inhibitors, the drugs that turned metastatic melanoma from a death sentence into something survivable, did almost nothing here. In the KEYNOTE-199 trial, pembrolizumab produced an objective response in about 5 percent of men with treatment-refractory metastatic castration-resistant disease, and that was the measurable-disease cohort, the one with the best odds. Two Phase III programs that added the drug to standard enzalutamide therapy, KEYNOTE-641 and KEYNOTE-991, were both halted at interim analysis after crossing the boundary for futility. Adding it to docetaxel failed too, in KEYNOTE-921. Oncologists have a name for a tumor that behaves this way: immunologically cold.

Cold does not mean the immune system literally cannot see the tumor. It means the tumor gives it almost nothing to grab. A T cell will destroy a cancer cell if it can recognize one, and recognition runs through a surface structure called MHC class I, the molecular billboard that displays a cell’s internal contents for inspection. Take down the billboard and the tumor goes dark. Prostate cancer, it turns out, is very good at taking down the billboard, and the paper Eric Wagner’s group just published in Nature Biomedical Engineering is largely an account of how.

The mechanism is elegant in the way only something evolved to kill you can be. In the model, the tumor cells overproduce a protein called SPSB1, whose job is to degrade the MHC-I complex, and they crank up production through a quiet trick of RNA processing. The messenger RNA that codes for SPSB1 normally carries a long tail, a stretch of untranslated sequence at its 3’ end that acts as a brake on how much protein gets made. The cancer cells shorten that tail through a process called alternative polyadenylation. The shortened message is more stable, gets read more often, and the cell floods itself with SPSB1. More SPSB1, fewer billboards. Fewer billboards, and the immune system, along with any checkpoint drug meant to unleash it, has nothing to aim at.

So the team, working with collaborators led out of Duke, built a tool to grow the tail back. They call it the 3’UTR CRISPR/dCas13 engineering system, and its defining feature is that it cuts nothing. Conventional CRISPR is a pair of scissors. This is a clamp. A catalytically dead Cas13 protein is steered onto the exact stretch of the SPSB1 message where the trimming happens, and by parking there it blocks the cell from shortening the tail. The message stays long, the brake stays on, SPSB1 falls, MHC-I climbs back to the cell surface, and in the animals the tumors that had been invisible started drawing T cells again. Checkpoint therapy that had done nothing became effective. The researchers reported more immune infiltration and more cancer-cell killing, and in the assays they ran, no detectable off-target editing, which is the first objection anyone raises about a technology with CRISPR in the name.

The tool worked, and the RNA-engineering piece is the part worth being impressed by. Re-lengthening a specific mRNA on purpose, inside a living animal, to dial down one protein, is a capability the field has wanted for years. The money came from the National Cancer Institute, with pilot funding from the Wilmot and Roswell Park cancer institutes. The taxpayer, in other words, paid for the tool that might one day rescue the checkpoint drugs a pharmaceutical company spent a fortune failing to make work.

This is not a treatment yet, and two things stand between it and a patient. Wagner said the quiet part in the press materials: “If we can hit it with immunotherapy and another synergistic drug that pumps up the immune response, we could potentially cure it.” Cure is a large word to hang on a mouse. The first obstacle is that a controlled result in a mouse tumor is not a result in a man, a distinction oncology has learned and relearned for fifty years. The second is delivery, and it is the one that matters: getting a dCas13 clamp into enough cells of an actual prostate tumor to move the needle is a problem this paper does not solve, because it is nowhere in it. The study proves a mechanism. It does not prove a medicine.

Wagner’s lab has already aimed the same tool at pancreatic cancer, another famously cold tumor, which tells you where this is headed before it gets anywhere near a person. The honest version is smaller than the headline and more interesting than the hype. For ten years the field kept giving checkpoint inhibitors to prostate-cancer patients and kept watching them fail, and it took a lab picking apart the mechanics of an RNA tail to say, at the level of a single sabotaging protein, why. In the mouse, the drug finally had a target again. The tumor had spent the whole time quietly unplugging the one thing the drug was built to find.

Sources

  1. Nature Biomedical Engineering – Programmable mRNA 3’UTR engineering restores MHC-I and overcomes immune evasion in prostate cancer (2026)
  2. University of Rochester Medical Center – Making Immune Therapy More Effective in Prostate Cancer
  3. ScienceDaily – CRISPR makes prostate cancer vulnerable to immunotherapy
  4. Journal of Clinical Oncology – Pembrolizumab for Treatment-Refractory mCRPC: KEYNOTE-199 (2020)
  5. AACR Cancer Discovery – PD-1 Blockade Falls Short (Repeatedly) in Prostate Cancer (2023)
  6. Journal of Clinical Oncology – Pembrolizumab Plus Docetaxel: KEYNOTE-921 (2024)